Touchstone Semi?

Not even close. Try 25% of that.

I tell the FAEs what I'm looking for and they come back with a selection of parts, usually with prices. It's impossible to design without the right prices or at least really close.

Different strokes, I suppose. I can't imagine buying and placing a billion transistors on each board.

You have to trim resistors to get closer than 10% or 20%? Boggle! We use a lot of 5% resistors, though I don't use a different PN if a 1% is already on the board. A feeder isn't worth the difference between

won't buy anything he can't understand".

Are you *EVER* going to make sense, Slowman?

I learned the lesson that this is not so early in my career. A raw materials plant in Asia had an explosion. Suddenly certain single-sourced high-density caps became unobtanium for many month. Those of us who had used them got stung.

But of course it all paled when we heard that several workers lost their lives in that explosion.

I subscribe to a different philosophy, that of minimizing risk exposure as far as reasonably possible. And I find that it usually is possible.

The only time I have seen 25% as with custom ferrite parts. But I do not design stuff where we need 10 million/year, it's probably different in that market.

When Findchips lists the LM324 at $0.05 then it is usually impossible to get it for 25% of that. Unless one settles for a knock-off of dubious pedigree.

Then you guys should look into a system that one of my old clients in Asia used. Early days of the Internet, where people still printed stuff out a lot. We designed and after an idea for a subcircuit looked like "that's it" we trudged over to a hallway where they had the whole long wall plastered with tractor-printer paper and prices on there. This wall was updated about every week and it was "Asia prices", much lower than what I was used to. Came very close to the real deal. I asked the VP Engineering where they got that data but he was not at liberty to reveal, only said "We have our sources for this stuff".

I did or, rather, the contract mfg does :-)

But it does not work with domestic placement costs.

I wrote "non-trimmed". We used only catalog resistors, where 1% was needed we used that, and where it wasn't I loosened the spec accordingly. I once received a request from the Chinese contract assembler of a client, asking which resistors could be eased to 10% and if possible maybe some to 20%. So I went through this old design of mine and gave them a list. This activity cost my client 1/4h of billed consulting time and saved them the equivalent of a nice car. Every year. The only way to calculate that kind of ROI is in dB or orders of magnitude.

If you used one, your production was shut down. If you used a dozen, so what?

Sure. Prospective. The floods in Thailand and various earthquakes were no picnic, either. We scramble to make deliveries. They scramble for their lives.

I play by their rules. It works.

Don't know about an LM324. I haven't used one of them for forty years. The millions pieces pricing of semiconductors is often a quarter to a half of what you see in price lists. There is no set ratio, though, so quotes are needed for every part. It can vary by the day you ask, too. ;-)

Spot market prices, no doubt. These are hardly useful for contracts that go for years. Getting the prices right is the difference between profit and loss. I like profits better.

expensive

the

No, they don't. They don't fit. ;-)

Simple. If you have one ciruit block where a redesign needs to be done it can often be done in days, plus the usual relayout and fab time. If there are dozens of cases and essentially the whole board is toast and needs a redesign, very different ballgame.

But it's your choice. As I said, my philosophy in that respect seems to differ from yours.

smaller

Time to check it out, I'd say. The only way to beat its price is with form-fit-fucntion compatible alternative that are in essence like the same thing.

My experience is different. I have never seen a semiconductor go to 25% of its 15k price. I've had designs where stuff goes into serious six-digits per year and hardcore purchasing negotiators take over. It tends to taper off gradually, in the end they are haggling over the last

Definitely not.

They consistently made profits because this price info was correct. They had to pay for it but it sure beats waiting for 20-30 phone calls from vendors each day, calls I really don't have the time for.

it.

No, the difference is really very small. If it was a trivial part to begin with, there is no need to do much at all. If it's a more complex part, it's a complete redesign anyway, so another isn't going to change much.

No, it really isn't my choice.

smaller

me

I haven't had the reason to use a POS like that in forty years. In case you hadn't noticed, there are *far* better parts out there now. ;-)

If you've already designed in an expensive part, you've already lost the game. Of course you didn't know that when you designed it in because you didn't know the price. Haggling after the design is complete is the *wrong* way to do it. Horse ==> Cart

Have to be.

The price is correct, today. Tomorrow... The vendors themselves don't know that the prices will be until you ask. They *will* give two people different answers on the same day.

umps the never-was.

Apparently they are now part of Carl Zeiss Electron Microscopy Ltd.

There seem to be about 100 employees on site - there were 400 when we parte d company back in 1991 but the machine shop - lathes, milling machines and the experienced machinists who worked them - all got floated off as a separ ate company sometime in the 1990's.

won't buy anything he can't understand".

Not to you. I'm aiming further up-market.

won't buy anything he can't understand".

You've struck mud. You really are dumb as a stump.

won't buy anything he can't understand".

Usenet is like that. You post stuff and some gormless idiot can't understand it and complains that you haven't dumbed it down to his level.

and complains that you haven't dumbed it down to his level.

I like that, a Brit slander new to me, but a perfect descriptor for the occasion >:-} ...Jim Thompson

OK, I just followed what is needed to make the output swing upwards. The top Sziklai pair must conduct, so it's base should go down. For this, the driver NPN should conduct more, so the emitter follower before it should conduct. For this the right half of the differential pair should start to conduct. Therefore its base should swing down - but it is marked "non-inverting" and it should swing up in order to make the output swing up.

But you're absolutely right, I didn't check what the lower Sziklai pair does. Funny, I wonder who drew the schematic, and how much it resembles innards of the actual IC.

Regards, Mikko

It makes more sense for the upper pair to be a simple Darlington. That's consistant with the three junctions in the bias string.

rote:

Yes, that's probably the simplest explanation about how it is implemented in real life. Someone has simply drawn one arrow in a wrong location.

Anyone who has seen the totem pole output of a 74 TTL gate should have recognized the mistake right away. But the additional PNP's obviously confuse the automated mental image recognition.

Regards, Mikko

worries about discontinuing parts, face it, they're still making products pushing 45 years of legacy!

will have one too, rest assured.

The stray capacitance is largely constant and so in my app is not important. This is a VCO, not expected to be an exact frequency for an exact voltage. The VCO control will adjust out any offset in the center frequency.

If I understand their design the ramp up and ramp down times are equal, producing a square wave. I believe the drawing shows a constant current and a switched inverter.

Yes, I have looked at the data sheet, but they don't show any examples of operation near 240 kHz, not that it matters much. They do say the device works at higher speeds than this. The FAE was nice enough to test an eval board he has.

My only concern is how sensitive the VCO control input is to the applied voltage. I want something that isn't too sensitive. I just need to be able to bring it to the set point with variations in temperature and components. I asked the FAE a couple of times, but we just didn't connect somehow on what I was concerned about. So a little experimentation will be in order once I get a circuit built I can test.

I won't be working on this for a bit. I just got a large board order. So I need to work to improve the test fixture so I can get the boards tested and out the door. Maybe in a couple of months I'll have more time for the very low power VCO.